Re: [PATCH v3 1/3] xen/blkfront: read response from backend only once

[Oleksandr Andrushchenko]

On 02.08.21 22:26, Julien Grall wrote:
> Hi,
>
> On 02/08/2021 15:06, Oleksandr Andrushchenko wrote:
>> On 30.07.21 13:38, Juergen Gross wrote:
>>> In order to avoid problems in case the backend is modifying a response
>>> on the ring page while the frontend has already seen it, just read the
>>> response into a local buffer in one go and then operate on that buffer
>>> only.
>>>
>>> Signed-off-by: Juergen Gross <jgross@xxxxxxxx>
>>> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
>>> Acked-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
>>> ---
>>>    drivers/block/xen-blkfront.c | 35 ++++++++++++++++++-----------------
>>>    1 file changed, 18 insertions(+), 17 deletions(-)
>>>
>>> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
>>> index d83fee21f6c5..15e840287734 100644
>>> --- a/drivers/block/xen-blkfront.c
>>> +++ b/drivers/block/xen-blkfront.c
>>> @@ -1496,7 +1496,7 @@ static bool blkif_completion(unsigned long *id,
>>>    static irqreturn_t blkif_interrupt(int irq, void *dev_id)
>>>    {
>>>        struct request *req;
>>> -    struct blkif_response *bret;
>>> +    struct blkif_response bret;
>>>        RING_IDX i, rp;
>>>        unsigned long flags;
>>>        struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)dev_id;
>>> @@ -1513,8 +1513,9 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
>>>        for (i = rinfo->ring.rsp_cons; i != rp; i++) {
>>>            unsigned long id;
>>>    -        bret = RING_GET_RESPONSE(&rinfo->ring, i);
>>> -        id   = bret->id;
>>> +        RING_COPY_RESPONSE(&rinfo->ring, i, &bret);
>>
>> As per my understanding copying is still not an atomic operation as the request/response
>>
>> are multi-byte structures in general. IOW, what prevents the backend from modifying the ring while
>>
>> we are copying?
>
> Nothing and, I believe, you are never going to be able to ensure atomicity with large structure (at least between entity that doesn't trust each other).
>
> However, what you can do is copying the response once, check that it is consistent and then use it. If it is not consistent, then you can report an error.
>
> This is better than what's currently in tree. IOW we may have multiple read so the code is prone to TOCTOU.

Agree,

Thanks

>
> Cheers,
>